The input pattern given to the string_match() method is not a regex
pattern and we cannot use it as that, because all the special
characters (e.g. '.') would not be treated correctly.
Also, the prefix matching with the wildcard at the end of the string
needs to be converted to a proper regex wildcard, i.e. '.*'.
This logic also adds support for suffix matching, with a wildcard at
the beginning of the string, e.g. as the ones used for the wwan kernel
device name matching rules (i.e. '*MBIM').
Unit tests are added to cover all cases, most of these tests were
failing without the fixes implemented here.
Different options are set as `feature`, also every possible plugin,
but this may cause confusion.
All options set as `feature` but plugins have been made `boolean` to
avoid further confusions.
`help2man` is also installed in the CI image because is required to
build different libraries.
We keep the 'WHITELIST-ONLY' filter type name still as an option in
--filter-policy=[POLICY], but deprecated and with the better
'ALLOWLIST-ONLY' replacement suggested from now on.
We let protocol implementations provide the uplink and downlink speeds
agreed with the network as part of the connect result, and we then move
them to the bearer stats in DBus.
Until now we just reported any registration status change as PS, and
that was fine because we wouldn't need to have any logic on top to
build a consolidated registration state.
But now we need to know whether the reported registration status is
for a packet domain (e.g. PS/EPS/5GS) in order to correctly update the
PacketServiceStatus property in DBus.
So, monitor the packet service status reported by the modem, and
decide which domain registration we're updating every time.
If we're registered in either PS, EPS or 5GS, we can assume packet
domain service is attached.
If we're only registered in CS, packet domain is detached.
This change relies on protocol implementations to properly report
separate CS/PS/EPS/5GS domain registration states.
Without setting memory to NULL it is possible that ctx->list
point to an unitialized pointer and trig a segfault when we
free it (for example when the firmware list is not supported)
Signed-off-by: Frederic Martinsons <frederic.martinsons@sigfox.com>
It stores the epoch timestamp of the current bearer session start.
If there is no connected bearer, it is set to 0 and not displayed
in mmcli output.
Signed-off-by: Frederic Martinsons <frederic.martinsons@sigfox.com>
Includes updates by Aleksander Morgado to fix coding style issues.
The values exposed in the Signal interface must not be cleared every
time polling is disabled, because the user may have also enabled
threshold based loading. And viceversa; if the user disables threshold
based polling, we should not unconditionally clear the values as
polling may still be enabled.
We setup a common Private context associated to the interface, and we
keep the current state there, which is also in sync with the DBus
interface.
We will only clear the signal values if both polling-based and
threshold-based setups are disabled.
Following the same reasoning, the mm_iface_modem_signal_update()
method used by implementations to report new signal quality details
is updated so that it's a no-op if no polling-based or threshold-based
setup has been enabled.
Also allow updating the per-access technology signal quality
information via indications.
Includes updates by Aleksander Morgado to fix coding style issues and
some other GTask related problems.
When changing the allowed modes using the "Register State Set"
request, the response would arrive once we have been registered in the
network. This is obviously not ideal, as we just want to know if the
mode preference was changed, we don't care if we're registered or not.
Instead of waiting for the response to arrive or to timeout, we now
also process incoming indications that arrive during the wait time,
and if any of them shows the preferred modes to be the same ones as we
just requested, we complete the operation right away.
In MBIM we use the same "Register State Set Request" for 2 different
things: configuring the allowed modes, and selecting the operator to
use (manual/automatic).
We need to keep track of which were the last things requested by the
user so that we don't overwrite previous user actions when we process
newer ones.
Use the new Preferred Data Classes field in the Register State v2
message in order to know if the modes requested in the Set message are
the expected ones or not.
Based on an initial implementation by Som_SP <somashekhar.puttagangaiah@intel.com>
The ModemManager codebase has been ported to support Basic Connect v2
messages as defined in MBIMEx v2.0, so at this point we can now switch
to use it by default whenever possible.
We're going to enable the Microsoft extensions by default, and in
order to properly parse the notification messages, we'll need to know
which is the currently agreed MBIMEx version.
So, pass down the MbimDevice to every notification parser. Not all of
them will need it, so just for completeness.
Since the 'wwan' subsystem addition to the qcom-soc plugin, the rules
to say a modem is handled by this plugin are too broad, and the plugin
attempts to support any kind of device exposed by the WWAN subsystem,
not just those from Qualcomm SoCs.
Update the plugin to require a new ID_MM_QCOM_SOC udev tag that is set
in the plugin udev rules, given that these rules already check for the
expected SoC drivers (bam-dmux, ipa, qcom-q6v5-mss).
Aleksander Morgado
Daniele Palmas
Iñigo Martínez
Som_SP
Frederic Martinsons
Andrew Lassalle